The Internet of Things (IoT) enables real world objects to be integrated into a virtual world, where sensors, actuators, and other devices interact not only with human users, but also with each other and software agents on the Internet. One approach for making IoT data available to users is the use of Web service technologies, which can directly integrate IoT data and Web functionalities through the Internet. This integration of Web services with IoT has been defined as the Web of Things (WoT) [1
]. Furthermore, sensors in IoT environment have been miniaturized, integrating various communication functions, such as Bluetooth, ZigBee, Low-power WiFi, and GPS.
The Internet Engineering Task Force (IETF) has undertaken much standardization work related to WoT. For example, the IETF Constrained RESTful Environments (CoRE) Working Group (WG) has been creating standardizations for introducing the Web service paradigm into networks of smart objects. The CoRE WG has defined a REST-based Web transfer protocol, the Constrained Application Protocol (CoAP) [2
]. CoAP can make it easy to integrate physical devices with contents on the Web, while satisfying requirements, such as multicast support, low signaling overhead, and simplicity for constrained network environments. The devices in a constrained network environment generally tend to be embedded, and to require considerably less CPU processing, memory, and power supply capabilities than Internet devices. More specifically, the constrained node often have 8-bit microcontrollers with small amounts of ROM and RAM, while constrained networks such as IPv6 over Low-Power Wireless Personal Area Networks (6LoWPANs) often have high packet error rates and a typical throughput of 10 s of kbit/s. Therefore, the requirements of multicast support, low signaling overhead, and simplicity are extremely important in a WoT environment. In addition, the constrained networks in a WoT environment usually have a limitation in packet size, may exhibit a high degree of packet loss, and may have a substantial number of devices in sleep mode operation [2
]. In a duty-cycled network, sensor nodes perform four distinct operational states: transmit, receive, idle, and sleep. In sleep states, the sensor is shut down and a low-power timer is on to wake up the sensor at a later time [3
]. Therefore, it can consume only a tiny fraction of the energy consumed in the active mode.
The interaction model of CoAP is similar to the client/server model of HTTP. However, unlike HTTP, the end-points of the CoAP may act as both clients and servers [2
]. The architecture of CoAP is divided into two layers: message and request/response. CoAP supports reliable message transmissions by using stop-and-wait retransmissions with an exponential back-off mechanism to correct the order of packets and check duplicates [1
]. CoAP can benefit various IoT applications, such as ubiquitous healthcare services, V2I/V2V automatic vehicle networks, home networks, automotive networks, automatic systems, industrial networks, interactive toys, and remote meters.
One limitation of CoAP is that it does not directly address the mobility requirements for mobile constrained nodes in WoT environments. CoAP has been designed for Machine to Machine (M2M) applications, such as smart energy and building automation [5
]. In previous WoT applications, WoT researchers assumed that most sensor nodes do not have movement. Therefore, the sensor mobility has not been considered in a WoT network environment. A sensor node, however, can have mobility. For example, in a vehicle monitoring system, the vehicle can move into different IP domains. In the ubiquitous network, the characteristics of the wireless network or sensor movement can change the wireless connection between the sensor and wireless access router. In WoT environment, the sensor provides the sensing resource and acts as Web server. In this paper, CoAP node indicates its sensor, which can provides the sensing resource, responses the request of another client, and is equipped CoAP protocol in the constrained network environment. As a CoAP sensor moves around different wireless networks, its IP connectivity may be disrupted, which may result in the loss of important sensing data, or delay of time-critical data. In CoAP, the IP address of CoAP node, is registered with the Domain Naming Server (DNS). The DNS configuration involves operation by humans as much as possible. If the CoAP server node moves between different IP domains, the client may not find the server location, i.e.
, IP address, if the human may not configure the changed IP address on DNS in time.
To prevent the previously described mobility problem of a Web server node, an existing mobility management protocol may be used. The IETF has developed various standard mobility management protocols. The mobility management for network layer, Mobile IPv4/v6 (MIPv4/v6) [6
] and its variants, including Fast Mobile IPv4/v6 (FMIPv4/v6) [7
], Hierarchical Mobile IPv4/v6 (HMIPv4/v6) [8
], and Proxy Mobile IPv4/v6 (PMIPv4/v6) [9
] were developed. The transport layer uses TCP migrate and the mobile Stream Control Transmission Protocol [10
]. For the application layer, SIP-based approaches [11
] have been proposed to manage mobility in next-generation wireless networks.
Unfortunately, most standard mobility management protocols add high signaling overhead from tunneling and binding operations, and are quite complex, incurring high processing and energy consumption. Additionally, most standard mobility management protocols require the modification of the network infrastructure such as Internet access router and mobile nodes. Furthermore, these standard mobility management protocols do not address the characteristics of a constrained IP network, such as limitations in packet size, high packet loss ratio, and sleep mode operation. Therefore, the protocols mentioned previously may not be suitable for mobility management in a WoT environment with constrained device and network characteristics; for example, with low processing and energy constraints, or in sleep mode operation.
With regard to the objective functionality of mobility management, the objective of a WoT environment differs from that of existing IETF mobility management protocols. More specifically, in conventional IETF mobility management protocols, the objective of mobility management is to enable a mobile node to initiate a session and be provided with an application service in a seamless manner during an IP handover. In a WoT environment, however, the objective is to enable a mobile sensing node to timely send measured data to a remote client whenever the client requests it. Therefore, a WoT environment needs a novel mobility management protocol that can satisfy the previously described objective, considering constraints on processing capability, energy consumption, and other characteristics, such as sleeping mode operation.
Jara et al.
presented a lightweight Mobile IPv6 with IPSec, which is aware of the requirements of the IoT and analyzes the efficiency and security adapted to IoT-devices capabilities [12
]. The authors proposed the lightweight Mobile IPv6, which does not execute the route optimization and return routability of the original MobileIPv6, to be integrated into constrained devices with a low capacity in terms of memory and communication capabilities. Additionally, the authors investigated the requirements for supporting the mobility management in IoT environment [12
]: global identifiers, IPv6-based protocol, communication costs, packet encapsulation, and movement detection. In the lightweight Mobile IPv6, the home agent and foreign agent play a role as middle agent in order to deliver the ingoing packet, i.e.
, control packet and real packet to the mobile node or corresponding node. As a result, the load of middle agent can be dramatically increased when the number of mobile nodes increases and the triangular routing problem can be incurred. Hence, the process of control packet and real packet may be separated. This lightweight MobileIPv6 does not consider the sleep mode operation of IoT devices and requires the modification of infrastructure such as home agent and foreign agent because it is based on Mobile IPv6.
Sungmin et al.
proposed the Sensor Networks for an All-IP World (SNAIL) based on MARIO [14
]. In this research, the sensor is composed of PAN coordinator, static node, partner node, mobile node, and gateway. SNIL uses the ancestral concept to perform the handover. More specifically, the mobile node retrieves the domain information of next static node, i.e.
, node ID and IP address, through the partner node before the mobile node performs the handover. After that, in the next domain, the mobile node performs the binding update with next domain information. As a result, the handover delay can be reduced. However, as the mobile node does not move into pre-defined location of next static node, the handover delay and packet loss can be large. Also, a PAN coordinator may always manage and update the information of near sensor. It can occur a large signaling overhead in the network domain.
Jara et al.
presented a protocol to carry out inter-WSN mobility inside of the architecture that has been defined at a hospital [15
]. It can decrease the number of interchanged messages of mobile nodes when the mobile nodes move within pre-defined regions. However, it is not suitable for IoT global mobility protocol because this mobility protocol cannot support the global mobility and the modification of network infrastructure is required. Kai et al.
presented the Care-of Address Pool for Hierarchical MIPv6 (CoAP-HMIPv6) to reduce the handover latency by reducing influence caused by the DAD procedure [16
]. However authors have not considered the mobile network with the constrained resource. Gligoric et al
. have proposed the Open Mobile Alliance device management protocol for reliable Device Management (OMA-DM) and have analyzed and compared the efficient XML interchange (EXI), CoRE Link format, and protobuf for efficient message format [17
]. The authors proposed EXI is efficient as the payload format in use of CoAP.
Berguiga et al.
presented a mobility management scheme for 6LoWPAN sensor nodes [18
]. The authors proposed the fast handover proxy mobile IPv6 for sensor network (FPMIPv6 S) protocol, an improved version of the Proxy Mobile IPv6 (PMIPv6) protocol, to reduce the number of messages exchanged and the handover latency. However, they did not consider the complexity of FMIPv6, with respect to CPU processing overhead and energy consumption.
Ganz et al.
presented a resource mobility scheme for service continuity in an IoT environment [19
]. They proposed a resource mobility scheme using two operating modes, caching and tunneling, to enable applications to access the sensory data when a resource becomes temporarily unavailable. The sensor gateway caches the measured data, and transmits the data in response to a service provider’s request instead of the sensor. The tunneling method reduces the amount of packet loss during the handover of a sensor by creating a tunnel between the sensor gateways. However, as both sensor gateway and sensor itself can move between different wireless networks, the connectivity might be disrupted during their movement.
In summary, most current mobility management protocols may not be suitable for supporting the mobility of CoAP sensor nodes in WoT environments because the sensor nodes in such an environment generally have constrained CPU processing power and memory capacities and they must have low energy consumption. They have other characteristics such as sleep mode operation and a constrained network of wireless sensor networks. Current mobility management standards of the IETF have not addressed these constraints on the design of mobility management architecture and protocols.
In this article, we propose the CoAP-based Mobility Management Protocol (CoMP), which can provide mobility management for mobile CoAP sensor nodes. Because CoMP uses a separate location management function, which is based on CoAP, low signaling overhead can be obtained due to simplicity of the mobility management architecture. The tunneling scheme is not used for architectural simplicity. CoMP enables the IP addresses of mobile CoAP sensor nodes to be kept track of, allowing monitored sensing data to be reliably delivered to Web clients using both HTTP and CoAP. To the best of our knowledge, there have been no previous research attempts at providing direct IP mobility functionality to mobile CoAP nodes. Compared with other related works, the originality of our approach may be summarized as follows:
Instead of designing new signaling protocols for mobility management, CoMP employs the IETF standard CoAP protocol for mobility management in an application layer, without changing the lower layer. This achieves the simple seamless connectivity of wireless constrained sensor node without the modification of the existing network infrastructure.
CoAP messages and methods are extended to implement the mobility management functions of a mobile CoAP node, which imparts not only simplicity in the mobility management architecture, but also has significantly low signaling overhead, compared to other protocols, such as MIPv4/v6 and its variants.
In the existing IETF MIPv6 mobility management protocol, a bi-directional tunnel scheme has been used for transparent handover operation. Instead of a bi directional tunnel, CoMP uses two modes of operation, holding and binding, for fast and reliable data transmission.
The contributions of our research are as follows:
The detailed architecture and functions of CoMP have been designed for mobility management. A separate location management function to support CoAP service mobility has been designed.
The sleep mode operation of sensor node in CoMP is considered to provide reliable service.
Detailed signaling procedure and an address management method were designed for supporting seamless connectivity and reliable transmission.
To enhance interoperability, we extended CoAP; more specifically, CoAP messages and methods were extended to exchange messages for managing IP addresses between CoAP nodes.
The remainder of this paper is organized as follows. In Section 2
, we describe the overview and limitations of the CoAP standard. We also describe the comparison of CoMP with the existing standard mobility management protocols such as Mobile IP and SIP-MM. In Section 3
, we describe the architecture and message formats of the proposed CoMP. In Section 4
, we present a mathematical analysis of the proposed CoMP handover mechanism for a performance evaluation. In Section 5
, we describe the performance results of the proposed scheme. Finally, in Section 6
, we provide some concluding remarks regarding this research.